Kling 3.0 vs Kling 3.0 Omni: inputs, limits and price

Kling 3.0 and 3.0 Omni share 3–15 s clips up to 4K with native audio. Omni adds reference and edit videos; 3.0 adds Motion Control. Prices compared.

5 min readSume
All posts

Kling 3.0 and Kling 3.0 Omni make the same kind of clip: 3–15 seconds at 720P, 1080P or 4K, with multi-shot and native audio. The difference is what you can feed them. Omni also takes reference images and a video, either as a reference or as the clip to edit. On its text and image to video endpoints, plain Kling 3.0 takes a prompt, first and last frames and Elements, and it has Motion Control, which Omni lacks.

Every Kling fact below comes from Kling's own API docs and user guide, read on 2026-09-28 and listed under Sources; Kling's capability map was last updated on 2026-05-19. The Sume section uses Video generation.

How do Kling 3.0, 3.0 Omni, 3.0 Turbo and O1 compare?

Kling's capability map lists all four side by side. “Yes” means the map says supported for image to video, where the frame, element and reference rows live.

From Kling's video capability map, read 2026-09-28.
CapabilityKling 3.03.0 Omni3.0 TurboO1
InputsText, image, videoText, image, videoText, imageText, image, video
Length3–15 s3–15 s3–15 s3–10 s
Resolution720P, 1080P, 4K720P, 1080P, 4K720P, 1080P720P, 1080P
Multi-shotYesYesYesNo
Native audioYesYesYesNo
First/last frameYesYesNoYes
Element controlYesYesNoMulti-image elements only
Motion ControlYes, not at 4KNoNoNo
Video referenceNoYesNoYes

What is Kling Omni?

Kling's user guide says Kling VIDEO O1 was upgraded to VIDEO 3.0 Omni, and that Omni treats “the images, videos, elements, and text you upload” as prompts (3.0 Omni guide). On the API it has its own endpoint, POST /omni-video/kling-3.0-omni, whose contents array takes prompt, first_frame, last_frame, refer_image, feature_video, base_video and element. You name inputs in the prompt as @image_1 or @video_1 (Omni video generation).

  • One reference video at most: .mp4 or .mov, up to 200MB, 3 to 15.5 seconds.
  • With feature_video (a feature reference), multi_shot can only be true and audio only off.
  • base_video (the video to edit) allows no first or last frame, no multi-shot and no native audio.
  • settings.audio adds original, which keeps the reference video's sound.

How much do Kling 3.0 and 3.0 Omni cost?

Both are billed per second, from resource packages, at 720P / 1080P / 4K (video pricing):

  • Kling 3.0, no native audio: 0.6 / 0.8 / 3.0 Units ($0.084 / $0.112 / $0.42).
  • Kling 3.0, native audio without voice control: 0.9 / 1.2 / 3.0 Units ($0.126 / $0.168 / $0.42).
  • 3.0 Omni, no video input and no audio: 0.6 / 0.8 / 3.0 Units, the same as Kling 3.0.
  • 3.0 Omni, no video input with native audio: 0.8 / 1.0 / 3.0 Units ($0.112 / $0.14 / $0.42).
  • 3.0 Omni, with video input and no audio: 0.9 / 1.2 / 3.0 Units ($0.126 / $0.168 / $0.42).

What about Kling 3.0 Turbo and O1?

Kling 3.0 Turbo takes text or an image and stops at 1080P, with no first/last frame, Elements or 4K; its one price row is with native audio, 0.8 Units ($0.112) a second at 720P and 1.0 Unit ($0.14) at 1080P. Kling O1 is the Omni predecessor: 3–10 seconds, no multi-shot, no native audio, and single-shot clips only at 5 or 10 seconds, at 0.6–1.2 Units a second depending on video input and resolution (capability map, video pricing).

Which Kling model can I call through Sume?

Only Kling 3.0: Sume's catalog today lists kling-3 on POST /v1/videos, with no reference inputs, and Omni, Turbo and O1 are not in it (Video generation). Kling 3.0 API covers that request and its price. If you wanted Omni for reference videos, Sume's docs name the Seedance 2.x models, Wan 3.0, MiniMax H3 and H3 Max, and Gemini Omni Flash 1.1 as the models that take them. Kling 3.0 Motion Control is a separate Sume route, POST /v1/kling/3.0/motion-control (API reference); see Motion control API.

Sources

Related posts

More in Models

All Models posts

Written by Sume